拟万有引力定律三维激光点云数据精简方法

Three-dimensional laser point cloud reduction method based on imitating law of universal gravitation

为对海量三维激光点云数据进行精简,提出一种拟万有引力定律的点云数据精简方法。该方法将万有引力定律中各点质量替换为点的切向与径向联合曲率表征参数,通过求解点云中各点之间的引力分布,实现点云数据平坦与特征区域的划分,完成点云数据中特征点的提取与保护。将非特征区域点大比例均匀采样,再与完整的特征点融合形成精简后点云数据,将其与原始点云进行对比,结果表明:该精简方法在有效保留特征区域的基础上可以大比例精简点云,同时精简前后点云点距离误差较小。在总精简比94%时,最大点间距离偏差值为0.1 mm,且高偏差值点均位于非特征区域。

In order to reduce the massive 3 D laser point cloud data, a method based on imitating law of universal gravitation is proposed in this paper. In this method, the point mass in law of universal gravitation was replaced with the joint curvature representation parameter which is a combination of the point curvatures from tangential and radial directions. By presenting the gravitational distribution between each point, the point cloud data was divided into flat region and feature region, and then the extraction and protection of feature points in cloud data were achieved. The points in non-feature region were uniformly sampled in large proportion and then merged with feature points to form the final refined point cloud data. The refined point cloud was then compared with the original one. The experimental results show that the method can effectively preserve the feature region and effectively reduce point cloud in large proportion while the distance deviation between refined and original data is small. When the total simplification is 94%,the max distance deviation between points is 0.1 mm, and the points with high value of deviation are all located in the non-feature region.